Fuzzy pairwise Markov chain to segment correlated noisy data

This paper deals with an image segmentation process using a new fuzzy Markov model, which characterizes the imprecision of the hidden data and the correlation of the observed data. We propose to extend a recent pairwise Markov chain model (PMC [W. Pieczynski, Pairwise Markov chains, IEEE Trans. Pattern Anal. Mach. Intell. 25 (5) (2003) 634-639]) to a fuzzy context, allowing us to treat a spatial correlated noise between neighboring observations. The new algorithm, called fuzzy pairwise Markov chain (FPMC), requires a more specific methodology in order to compute the posterior density related to the hidden field. We validate our approach through experiments performed on synthetic and real images.